Mimir

Mimir Systems delivers an AI-driven tool for discovering and qualifying advanced materials, primarily aimed at teams engaged in materials research or those working with material development as part of engineering programs.

The system works by building a proprietary representation of materials, then triangulating from real-world data to create candidate materials within a searchable vector space. It uses a three-step approach: a proprietary representation of chemistry, structure and process; triangulation from surrounding experiments; and learning fastest from failure to prune the design space and guide progress.

Core capabilities include answering literature questions with reported properties and provenance, providing a live corpus with deep coverage, and offering on-premise and sensitive-data-ready deployment options. It emphasizes traceability, access for engineers, and evaluation against a user’s own corpus while maintaining a link to source documents.

Use cases

• Researchers solve materials problems faster by querying literature and testing ideas in a managed workflow.
• Teams without materials expertise can have discovery conducted as a service.